Expedia MCP
One of the more interesting shifts happening in AI right now is around discovery.
For years, the internet has largely worked the same way: search > scroll > compare > click. However, increasingly (note: still low single digit% usage in many sectors & SERPs are going nowhere anytime soon); people are starting to utilise LLMs within their customer reearch > buy journey.
Something like: “Find me a luxury hotel in Rome next weekend under £300 with a pool” in ChatGPT.
…Instead of paid + organic SERPs; they receive recommendations, live pricing, availability & booking options directly inside a conversation string.
What’s interesting technically is how quickly the ecosystem enabling this is forming.
Expedia’s integration into conversational AI is a good example. Underneath it sits structured inventory, APIs + emerging standards such as MCP (Model Context Protocol), which allows AI systems to securely discover tools and query live data in a more standardised way.
At a simplified level, the interaction increasingly looks less like search:
search(query=”rome hotels”)
…and more like orchestration:
mcp.call(“expedia.search_hotels”, { location: “Rome”, dates: “next weekend”, pool: true, budget_gbp: 300 })
AI is moving from generating answers toward coordinating systems. This shifts the interface layer between customers and brands. Increasingly, businesses will need to ensure their data is: machine readable, structured cleanly, accessible to AI systems & differentiated enough (think Google’s E-E-A-T yet for LLM’s) to become the recommended result inside AI-driven workflows. Interesting times ahead :-)